Meat and fish are usually cooked before being consumed. Cooking methods that induce high temperatures and a direct exposure to a hot surface, such as grilling or frying, or to direct flame, for example, barbecuing, are discussed in relation to carcinogenesis. Results from several epidemiological studies show associations between meat prepared at high temperatures and the risk of several types of cancerReference Nowell, Coles, Sinha, MacLeod, Luke Ratnasinghe, Stotts, Kadlubar, Ambrosone and Lang1–Reference Sinha, Kulldorff, Swanson, Curtin, Brownson and Alavanja5. One of the reasons for the increased risk is thought to be the heat-dependent formation of heterocyclic aromatic amines (HCA). When meat is cooked at temperatures over 130°C, for example, when frying, barbecuing or grilling, these compounds are formed from amino acids, creatinine and sugarReference Sugimura6. The amount of HCA production depends mainly on cooking method, temperature and the type of meat, with amounts ranging in most studies from 1 to 80 ng/g meat for 2-amino-1-methyl-6-phenylimidazo[4,5b]pyridine (PhIP), the most abundant HCA in the human diet, followed by 2-amino-3,8-dimethyl-3H-imidazo[4,5f]quinoxaline (MeIQx), with usual amounts up to 6 ng/g meat and 2-amino-3,4,8-trimethyl-3H-imidazo[4,5f]quinoxaline (DiMeIQx), with usually up to 1 ng/g meatReference Skog, Johansson and Jagerstad7. Recent epidemiological studies have shown associations between the estimated intake of HCA from diet and the risk of colorectalReference Nowell, Coles, Sinha, MacLeod, Luke Ratnasinghe, Stotts, Kadlubar, Ambrosone and Lang1, Reference Butler, Sinha, Millikan, Martin, Newman, Gammon, Ammerman and Sandler8, breastReference Sinha, Gustafson, Kulldorff, Wen, Cerhan and Zheng4 and prostateReference Cross, Peters, Kirsh, Andriole, Reding, Hayes and Sinha9 cancer, although other studies did not observe positive associationsReference Augustsson, Skog, Jagerstad, Dickman and Steineck10, Reference Delfino, Sinha and Smith11.
We examined the distribution of HCA intake from meat in a German cohort of middle-aged men and women. Previous studies reported a wide range of the daily HCA intake from as low as 77 ng/d in a Swedish studyReference Augustsson, Skog, Jagerstad, Dickman and Steineck10 to more than 1 μg/d in a US studyReference Keating and Bogen12. The intake of HCA in a pilot project has been described previouslyReference Rohrmann and Becker13. This new analysis was conducted in the entire European Prospective Investigation into Cancer and Nutrition (EPIC)-Heidelberg cohort.
Material and methods
EPIC is a prospective cohort study conducted in ten countries that started in the early 1990sReference Riboli, Hunt and Slimani14. In Heidelberg (Germany), 25 544 subjects, aged 35–65 years (women) and 40–65 years (men), were recruited for participation between 1994 and 1998. During recruitment, information on diet, lifestyle and health have been collected. All subjects are being contacted in approximately 2-year intervals to collect information on chronic disease status as well as diet and lifestyleReference Bergmann, Bussas and Boeing15. During the second follow-up (2001–3), 25 049 participants have been contacted. Of those, 86 % completed a 158-item FFQ that assessed food consumption during that past 12 months, which had previously been used in the baseline assessment (1994–8) (for details, see Bohlscheid-Thomas et al. Reference Bohlscheid-Thomas, Hoting, Boeing and Wahrendorf16). This FFQ included questions on meat consumption and preparation. Participants were asked how often they consumed sixteen types of meat (beef roast, including goulash, roulade; beef steak, filet or tenderloin; pork roast, including goulash; pork steak, schnitzel, cutlet, filet or tenderloin; hamburgers or meatballs; frying sausage; Wieners; bacon; liverloaf; fried chicken, turkey breast, turkey goulash) and which cooking methods they prefer for each type of meat (steaming or boiling, pan-frying, breading and frying, frying or broiling, grilling or barbecuing). Additionally, with the help of four pictures, subjects stated which degree of browning they favoured (lightly browned, moderately browned, strongly browned, extremely browned)Reference Augustsson, Skog, Jagerstad and Steineck17. If a subject indicated to vary between two or more cooking methods per food items these cooking methods were weighted equally. The same degree of browning was assumed for each cooking method used for a specific type of meat.
Total HCA concentration and concentration of the most abundant HCA PhIP, MeIQx and DiMeIQx were estimated using published data of their content in different types of meatReference Sinha, Rothman, Salmon, Knize, Brown, Swanson, Rhodes, Rossi, Felton and Levander18–Reference Skog, Steineck, Augustsson and Jagerstad21. HCA intake from steaming or boiling or from breading and frying was considered to be zeroReference Skog, Johansson and Jagerstad7. In addition to meat cooking, participants were asked about the use of meat drippings to prepare gravy. The intake of HCA from gravy was calculated by multiplying the amount of gravy with the HCA concentration in gravy of the corresponding meat item. This was automatically added to a specific meat item's HCA intake. By combining information on degree of browning, cooking method and the amount of meat intake, the mean daily dietary intake of HCA from meat was estimated. We also calculated HCA intake per MJ to take into account differences in energy intake that might contribute to differences in HCA intake. Because HCA intake was not normally distributed we computed medians and interquartile ranges of HCA intake and used the Wilcoxon test and Kruskal–Wallis test to compare the intake of different subgroups of our participants. All tests were two-sided; P values < 0·05 were considered to be statistically significant. EPIC-Heidelberg has been approved by the ethical committee of the Heidelberg University Medical School.
Results
The median intake of total HCA from meat was 30·6 ng/d (mean 69·4 ng/d), with PhIP contributing most to total HCA intake from meat (median 16·8 ng/d; mean 47·6 ng/d) (Table 1). Intake was highest from roast beef, followed by chicken or turkey, hamburgers or meatballs, and beef steak. Statistically significant differences in HCA intake were observed by sex, age, education, BMI and smoking status. Men had a higher HCA intake than women and smokers a higher intake than non-smokers (Table 2). Intake decreased with age and subjects with a higher educational level had a lower HCA intake than those with a lower educational level. Intake also differed by BMI, with more obese participants having a higher HCA intake from meat (Table 2). These differences were similar for HCA intake per MJ.
(Medians and interquartile ranges)
(Medians and interquartile ranges)
* Wilcoxon test.
† Age at second follow-up (2001–3).
‡ Kruskal–Wallis test.
Discussion
This is the first large European cohort study that attempts to assess the intake of HCA from meat using a detailed questionnaire on food intake and food preparation methods. We have previously estimated the intake of HCA in a smaller group of EPIC-Heidelberg participantsReference Rohrmann and Becker13, which revealed a higher intake of total HCA (median 103 ng/d) compared with the present investigation (median 30·6 ng/d). In a Swedish study that used a similar approach to assess HCA intake, median total HCA intake was 77 ng/d, which also included HCA intake from fishReference Augustsson, Skog, Jagerstad, Dickman and Steineck10. Similar amounts were calculated in a large Japanese studyReference Kobayashi, Hanaoka, Nishioka, Kataoka and Tsugane22 and slightly lower levels in a Singapore study (mean intake 49·95 ng/d)Reference Wong, Su, Knize, Koh and Seow23. In US studies, estimated HCA intake is generally higher (mean PhIP intake 78·1 ng/d; mean MeIQx intake 21·9 ng/d) than in European studiesReference Cantwell, Mittl, Curtin, Carroll, Potischman, Caporaso and Sinha24. This might be explained by larger portions of meat consumed in the USA than in Germany, but also by differences in cooking method preferences, for example, a preference for HCA-forming methods such as grilling in US cohorts and for non-HCA-forming methods such as boiling in Germany. Also, meat is usually consumed at a higher degree of browning in US cohortsReference Keating and Bogen12 than in our cohort, leading to a higher intake of HCA. In addition, different consumption habits contribute to the observed differences. ‘Roast beef, roulade and goulash’ contributed most to the intake of total HCA in our cohort; however, the contribution of roast beef to HCA intake was negligible in three US cohortsReference Byrne, Sinha, Platz, Giovannucci, Colditz, Hunter, Speizer and Willett25. The difference in HCA intake between the pilot studyReference Rohrmann and Becker13 and the present evaluation can, at least in part, be explained by some changes in questionnaire design. We added the possibility to mark the preparation method ‘boiling’ that does not contribute to HCA intake. Second, in contrast to the pilot study, we did not consider fish in the present study because HCA intake from fish varies widely depending on the type of fish and its preparationReference Skog, Augustsson, Steineck, Stenberg and Jagerstad20, Reference Gross and Gruter26. Third, HCA intake differed between subgroups mainly due to higher meat consumption, for example, in men, but also due to the preference of cooking methods or degrees of browning. Men tended to consume meat darker than women and younger participants tended to prepare meat more often by frying or grilling than older participants (data not shown).
In conclusion, we estimated an HCA intake from meat that was lower than observed in previous studies in Europe or the USA. Statistically significant differences were seen for age, sex, BMI, education and smoking status. EPIC-Heidelberg offers the opportunity to examine the association of meat consumption and HCA intake from meat with cancer risk in a prospective way. Further, future studies can take into account potential confounders, especially genetic variation in metabolic pathways of HCA as well as secondary plant products such as phenolic acids that are well known to have an impact on HCA metabolism and cancer risk.
Acknowledgements
The authors thank the volunteers who participated in EPIC-Heidelberg. This study was supported by the German Cancer Research Centre and grants from the Kurt-Eberhard-Bode-Stiftung and ECNIS (Environmental Cancer Risk, Nutrition and Individual Susceptibility), a network of excellence operating within the European Union 6th Framework Program, Priority 5: ‘Food Quality and Safety’ (Contract No 513943).
